The remote procedure call was cancelled.

[infopath]

In our organization we have 1000 servers that are domain controllers (one
server in each branch). All the servers are connected to central HUB (four
servers). Servers replicate once per 24 hours (because of large number of
servers i.e. optimization).
The servers are connected to the HUB over WAN connections (ADSL 1MBit/s
download; 128KBit/s upload). Every branch makes VPN connection (over the WAN)
to the central router.

Sometimes happens some servers to go down (WAN problem or hardware issue).
After bringing up those servers we expirience inbound replication problems:
When we try to replicate (from HUB to the local server) the naming context
(with repadmin) dc=domain,dc=com we almost immediately get "The remote
procedure was cancelled". But naming contexts replicate normally.
In event viewer we found:
------------------
Event ID: 1232
Active Directory attempted to perform a remote procedure call (RPC) to the
following server. The call timed out and was cancelled.
...
Call Timeout (Mins):
5
-----------------

We increased the RPC timeout to 35 mins. When we try to replicate the first
naming context the effect is the same. Immediatelly checking event viewer we
see that a timeout of 35 was reached and the RPC call was cancelled?!?!

Acording to http://support.microsoft.com/kb/898060/en we applied patches,
but they sometimes work, sometimes do not work?

When capturing net traffic we don't see RPC Bind request on the replication
RPC process when trying to replicate the first naming context. When we
replicate the other contexts RPC traffic looks good - the replication
occures?!?

All the machines run Windows 2003 Std with SP1.

[Emmanuel Antony]

Hi,

I would like to know, is there any Bridgehead servers for replication?If
yes, is it manually selected or selected by KCC.
--
Emmanuel Antony

Post by infopath
In our organization we have 1000 servers that are domain controllers (one
server in each branch). All the servers are connected to central HUB (four
servers). Servers replicate once per 24 hours (because of large number of
servers i.e. optimization).
The servers are connected to the HUB over WAN connections (ADSL 1MBit/s
download; 128KBit/s upload). Every branch makes VPN connection (over the WAN)
to the central router.
Sometimes happens some servers to go down (WAN problem or hardware issue).
When we try to replicate (from HUB to the local server) the naming context
(with repadmin) dc=domain,dc=com we almost immediately get "The remote
procedure was cancelled". But naming contexts replicate normally.
------------------
Event ID: 1232
Active Directory attempted to perform a remote procedure call (RPC) to the
following server. The call timed out and was cancelled.
...
5
-----------------
We increased the RPC timeout to 35 mins. When we try to replicate the first
naming context the effect is the same. Immediatelly checking event viewer we
see that a timeout of 35 was reached and the RPC call was cancelled?!?!
Acording to http://support.microsoft.com/kb/898060/en we applied patches,
but they sometimes work, sometimes do not work?
When capturing net traffic we don't see RPC Bind request on the replication
RPC process when trying to replicate the first naming context. When we
replicate the other contexts RPC traffic looks good - the replication
occures?!?
All the machines run Windows 2003 Std with SP1.

[Erik Cheizoo]

How about filtering routers and/or firewalls.
RPC needs high ports >1024.
Do you also see errors regarding the endpoint mapper?

There is a TechNet article how you can limit NTDS replication to a specific
port.
There's another one how you can limit NTFrs to a specific port.
--
Kind regards,

Erik Cheizoo
eXcellence & Difference - we keep your business running
============================================
Always test in a non-production environment before implementing
Guidelines for posting: http://support.microsoft.com/?id=555375
============================================

Post by infopath
In our organization we have 1000 servers that are domain controllers (one
server in each branch). All the servers are connected to central HUB (four
servers). Servers replicate once per 24 hours (because of large number of
servers i.e. optimization).
The servers are connected to the HUB over WAN connections (ADSL 1MBit/s
download; 128KBit/s upload). Every branch makes VPN connection (over the WAN)
to the central router.
Sometimes happens some servers to go down (WAN problem or hardware issue).
When we try to replicate (from HUB to the local server) the naming context
(with repadmin) dc=domain,dc=com we almost immediately get "The remote
procedure was cancelled". But naming contexts replicate normally.
------------------
Event ID: 1232
Active Directory attempted to perform a remote procedure call (RPC) to the
following server. The call timed out and was cancelled.
...
5
-----------------
We increased the RPC timeout to 35 mins. When we try to replicate the first
naming context the effect is the same. Immediatelly checking event viewer we
see that a timeout of 35 was reached and the RPC call was cancelled?!?!
Acording to http://support.microsoft.com/kb/898060/en we applied patches,
but they sometimes work, sometimes do not work?
When capturing net traffic we don't see RPC Bind request on the replication
RPC process when trying to replicate the first naming context. When we
replicate the other contexts RPC traffic looks good - the replication
occures?!?
All the machines run Windows 2003 Std with SP1.

[infopath]

To: Emmanuel Antony:
The four servers in the HUB are bridgheads.
KCC is running on every server with interval of 10 hours.

==============================
To: Erik Cheizoo
There are NO errors that talk about "endpoint mapper"

[infopath]

Interesting: when I create a dummy computer account in the HUB I can
replicate it normally using repadmin:
repadmin /replsingleobj baddc.domain.com HUB_dc_GUID
"cn=dummyPC,cn=computers,dc=domain,dc=com"

But when trying to syn the first naming context: just the same - the remote
procedure was cancelled???